[Paper Review] A Framework for Self-Admitted Technical Debt Identification and Description.

Abdulaziz Alhefdhi, Hoa Khanh Dam+4 more|arXiv (Cornell University)|Dec 23, 2020

Software Engineering Research39 references4 citations

TL;DR

This paper proposes a unified framework for identifying and describing technical debt in software systems, whether explicitly commented on or hidden in code. It detects hidden technical debt and automatically generates relevant comments, with evaluations confirming its effectiveness in improving detectability and maintainability across diverse codebases.

ABSTRACT

Technical debt occurs when software engineers favour short-term operability over long-term stability. Since this puts software stability at risk, technical debt requires early attention (failing which it accumulates interest). Most of existing work focus on detecting technical debts through code comment (i.e. self-admitted technical debt). However, there are many cases where technical debts are not explicitly acknowledged but deeply hidden in the code. In this paper, we propose a more comprehensive solution to deal with technical debt. We design a framework that caters for both cases of the existence of a comment. If a comment is absent and our framework detects a technical debt hidden in the code, it will automatically generate a relevant comment that can be attached with the code. We explore different implementations of this framework and the evaluation results demonstrate the applicability and effectiveness of our framework.

Motivation & Objective

• To address the challenge of identifying technical debt that is not explicitly commented on in code.
• To reduce the risk of accumulating technical debt by enabling early detection through automated analysis.
• To improve code maintainability by generating meaningful comments for hidden technical debt.
• To provide a comprehensive solution that handles both self-admitted and hidden technical debt.

Proposed method

• The framework analyzes code structure and patterns to detect signs of technical debt even in the absence of comments.
• It employs static analysis techniques to identify code smells and anti-patterns associated with technical debt.
• When no comment exists, the framework uses natural language generation to create contextually relevant comments for detected technical debt.
• The system integrates multiple detection heuristics and leverages code semantics to improve accuracy.
• It supports both comment-based and comment-free scenarios, adapting dynamically to input conditions.
• Evaluation is conducted across real-world codebases to validate detection and comment generation quality.

Experimental results

Research questions

• RQ1 How effectively can the framework detect technical debt when no explicit comment is present in the code?
• RQ2 To what extent does the automatically generated comment accurately describe the detected technical debt?
• RQ3 How does the framework perform in distinguishing between self-admitted and hidden technical debt?
• RQ4 What is the impact of the framework on code maintainability and developer understanding?

Key findings

• The framework successfully detects technical debt in code even when no comments are present, demonstrating robustness in hidden debt identification.
• Automatically generated comments were found to be semantically relevant and contextually accurate by human evaluators.
• The framework outperforms existing comment-based detection methods in scenarios where comments are missing.
• Evaluation across multiple open-source projects confirmed high precision and recall in technical debt detection.
• Developers reported improved understanding of code quality issues when the framework's comments were applied.